AI Creatures

AI Creatures is an ambitious project that delves deep into the realms of genetic algorithms and reinforcement learning. Leveraging the pyBullet library for creating dynamic physics environments and working extensively with URDF files, I embarked on a journey to birth virtual "creatures" with a unique ability to evolve and adapt within the digital ecosystem. These creatures, initially rudimentary in form, undergo a transformative process where they breed, mutate, and strive for survival in their simulated habitat. By the culmination of the simulation, the once nascent creature emerges as a highly adapted and 'fitter' entity, thanks to its genetic evolution.

The fitness of these AI creatures is meticulously measured based on the distance they move within the simulation, a testament to their adaptability and progress.

Highlights:

• Harnessing pyBullet Physics Environment: I harnessed the power of the pyBullet physics environment, creating a dynamic world where virtual creatures could evolve, adapt, and learn.
• Robotics with URDFs: By applying robotics methodologies and working extensively with URDF files, I crafted virtual creatures within the digital habitat, endowing them with joints and motive capabilities.
• Genetic Algorithm Mastery: The core of the project lies in the development of a genetic algorithm that evaluates, breeds, and mutates generations of creatures, paving the way for the emergence of progressively 'fitter' and more resilient entities.
• Parallel Processing Environments: To expedite the evolutionary process, I designed and implemented multiprocessing environments, allowing multiple generations to run simultaneously, accelerating the pace of adaptation.

Project Inspiration:

This project draws inspiration from Karl Sims' iconic "Evolving Virtual Creatures" experiment, seeking to replicate and extend the frontiers of artificial evolution in a virtual landscape.

---

More details & Features & Code files

Check out the GitHub repo: Github Repo